A Comparative Analysis of Direct Democracy, Two-Candidate Elections, and Three-Candidate Elections in an Experimental Environment

Abstract

This study explores the behavioral properties of political processes in a very simple environment. The processes are two-candidate elections, three-candidate elections, and large committees. These alternative political processes are implemented under laboratory experimental conditions in which the issues and underlying population preferences over issues are held constant. This allows the behavioral implications of the decision rule to be observed without the compounding complications caused by changing issues and attitudes. Such a setting provides a method for checking the predictive accuracy of spatial models and related game-theoretic models of candidate competition. If the models are sufficiently inaccurate in simple laboratory environments, then they might readily be rejected as applicable to the much more complicated, naturally occurring systems. In addition, the experimental design provides a comparison of selected aspects of behavior of these alternative processes.